compare

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow explicitly instructs the agent to fetch and ingest run histories from external experiment trackers (e.g., "wandb.Api(); api.run(...); history = run.history(...)" and "For neptune / mlflow, use their respective APIs"), which are user-generated third‑party data sources that the agent reads and uses to make comparison/decision outputs.